version 1.2, 2004/01/06 14:39:49
|
version 1.5, 2004/03/25 09:57:47
|
Line 2
|
Line 2
|
import xml.dom.minidom |
import xml.dom.minidom |
import os.path |
import os.path |
import re |
import re |
|
import urllib |
|
|
|
def getISO(): # very bad made has to be configured |
|
"""ISO""" |
|
try: |
|
f=file('/usr/local/mpiwg/Zope/lib/python/Products/OSA_system/iso639-1.inc','r').readlines() |
|
#f=file('OSA_system/iso639-1.inc','r').readlines() |
|
ret={} |
|
for lineraw in f: |
|
line=lineraw.encode('ascii','replace') |
|
value=string.split(line,'\t')[0].encode('ascii','replace') |
|
key=string.split(line,'\t')[1].encode('ascii','replace') |
|
ret[key]=value |
|
except: |
|
ret={} |
|
return ret |
|
|
|
def readArchimedesXML(folder): |
|
"""gib URL aus """ |
|
try: |
|
XML=urllib.urlopen("http://archimedes.mpiwg-berlin.mpg.de/cgi-bin/toc/toc.cgi?step=xmlcorpusmanifest").read() |
|
#print XML |
|
dom=xml.dom.minidom.parseString(XML) |
|
items=dom.getElementsByTagName('item') |
|
dict={} |
|
|
|
for item in items: |
|
#print item.attributes['dir'].value |
|
try: |
|
dict[item.attributes['dir'].value]=item.attributes['xml'].value |
|
#print item.attributes['dir'].value,item.attributes['text'].value |
|
except: |
|
"""nothing""" |
|
|
|
if dict.has_key(folder): |
|
return dict[folder] |
|
else: |
|
return "" |
|
except: |
|
return "" |
|
|
|
|
|
|
def getText(nodelist): |
def getText(nodelist): |
|
|
Line 45 def changeNodesInIndexMeta(path,node,sub
|
Line 87 def changeNodesInIndexMeta(path,node,sub
|
nodeOld=root.getElementsByTagName(subnode) |
nodeOld=root.getElementsByTagName(subnode) |
|
|
if nodeOld: # if yes -> delete it |
if nodeOld: # if yes -> delete it |
|
try: |
root.removeChild(nodeOld[0]).unlink() |
root.removeChild(nodeOld[0]).unlink() |
|
except: |
|
"""nothing""" |
|
|
namenode=dom.createElement(re.sub(r' ','-',subnode)) |
namenode=dom.createElement(re.sub(r' ','-',subnode)) |
namenodetext=dom.createTextNode(subnodes[subnode].decode("utf8")) |
namenodetext=dom.createTextNode(unicode(subnodes[subnode],"utf8")) |
namenode.appendChild(namenodetext) |
namenode.appendChild(namenodetext) |
root.appendChild(namenode) |
root.appendChild(namenode) |
|
|
Line 69 def changeNodesInIndexMeta(path,node,sub
|
Line 113 def changeNodesInIndexMeta(path,node,sub
|
for subnode in subnodes.keys(): |
for subnode in subnodes.keys(): |
|
|
namenode=dom.createElement(re.sub(r' ','-',subnode)) |
namenode=dom.createElement(re.sub(r' ','-',subnode)) |
namenodetext=dom.createTextNode(subnodes[subnode].decode("utf8")) |
namenodetext=dom.createTextNode(unicode(subnodes[subnode],"utf8")) |
namenode.appendChild(namenodetext) |
namenode.appendChild(namenodetext) |
nodeNew.appendChild(namenode) |
nodeNew.appendChild(namenode) |
|
|